Limits to Autocracy: An Analysis of China’s Renationalization 20170623...local government officials announced that the plan for Mr. Chen’s company to take control of the steelmaker

Limits to Autocracy: An Analysis of China’s Renationalization

Zhangkai Huang Lixing Li School of Economics and Management National School of Development Tsinghua University Peking University [email protected] [email protected]

Guangrong Ma Jun “QJ” Qian School of Finance Shanghai Advanced Institute of Finance Renmin University of China Shanghai Jiaotong University [email protected] [email protected]

This version: June 23, 2017

Abstract

We document large-scale reversal of privatization in China—local governments taking back shares in a quarter of previously privatized firms. Politicians who are not affiliated with any of the dominant political factions are more likely to waver under pressure and adopt renationalization. The failure to adhere to the privatization scheme reduces productivity and raises labor redundancy and firm leverage. The policy reversal casts doubt on the notion that autocracies have advantages in policy implementation: politicians without factional affiliation are disadvantaged in the promotion process and are more sensitive to unemployment pressure. JEL Classifications: G32, H11, P31, L22 Keywords: Renationalization, privatization, political faction, unemployment, productivity

1. Introduction

Are autocracies superior in implementing economic policies than democracies?

After all, since autocrats get their power through coercion, not election, it seems they

may be good at pushing economically sound policies that might be politically costly

under democracy. For example, Gary Becker states that:

“Visionary leaders can accomplish more in autocratic than democratic

governments because they need not heed legislative, judicial, or media constraints in

promoting their agenda…Visionaries in democracies can accomplish much

sometimes… However, their accomplishments are usually constrained by due process

that includes legislative, judicial, and interest group constraints.” (Becker 2010).

The experience of Taiwan, Singapore, South Korea, and more recently, China,

seems to support the view that, when ruled by visionary leaders, autocracies can be

efficient in promoting economic growth. As a comparison, democracies face far more

obstacles in policy formation.1 The idea of a benevolent autocrat henceforth appears

to be appealing, at least when he decides to adopt economic policies that have proved

to be effective.

In this paper we refute the notion that the authoritarian nature of an autocracy

puts it in advantage in adhering to good economic policies. We study China, perhaps

the most economically successful autocracy. We document large-scale

“renationalization”: local governments re-possessing ownership stakes in a quarter of

previously privatized firms. Thus although privatization improves firm firm efficiency

in general (see Megginson and Netter, 2001 for a comprehensive review), local

politicians in China fail to adhere to it. We find that such policy reversals are largely

driven by unemployment concerns, and that officials with NO affiliation with the

major political factions of the ruling Chinese Communist Party (CCP) are more likely

to reverse the course of privatization. The Princelings, those officials born into

political families, appear to be the strongest faction and are least likely to waiver

under pressure.

1 See Dinc and Gupta (2011), Dastidar, Fisman, and Khanna (2008).

Following renationalization, firms experienced reductions in profitability and

productivity, along with increases in labor redundancy and leverage. There are strong

political incentives behind the trade-off between efficiency and stability: we find that

factional ties boost the promotion likelihood of politicians, and that those officials

lacking strong patronage will find their promotion prospectus more sensitive to local

unemployment than to economic growth. They naturally care more about avoiding

social unrest than about making economic achievements.

Politicians in autocracies such as China are not constrained by an electorate, yet

their survival depends on their selectorate (Bueno de Mesquita et al, 2003). Although

privatization enhances efficiency in the long run, it may generate unemployment and

social instability in the short run. Politicians in charge of the privatization process risk

the probability of upsetting the top autocrat before reaping the benefits. This then

leads to policy distortions and policy reversals, and in turn generates economic

inefficiencies. While autocracies usually have disastrous economic outcomes under

incapable leaders, our results suggest that even a visionary autocrat may find his

vision fail to materialize.

We add to the literature on the relationship between political systems and

economic growth. Most advanced economies are democracies, and many countries

experienced democratization after their economy took off. Earlier modernization

theory argues that economic growth leads to democracy (Lipset 1959). Recently,

Acemoglu et al (2008) find that the correlation between income and democracy

disappears after controlling for country fixed effect, and there is no causal relationship

between income and democracy. Acemoglu et al (2017) further find that

democratizations in the past 50 years increase GDP per capita by about 20% in the

long run， and that democracy promotes economic growth by encouraging

investment, increasing schooling, inducing economic reforms, improving the

provision of public goods, and reducing social unrest. An interesting issue related to

these findings is that if a benevolent autocrat could mimic these policies and achieve

even more due to his authority. As Ferguson (2011) suggests, perhaps autocrats can

simply download the “killer apps” that have proved to be important for the success of

the West. This paper opens up the black box of how politicians implement economic

policies in an autocracy and our results show that the power struggle in a

non-democratic setting may actually hinder policy implementation, and that the

top-down system in an autocracy is not as efficient as it seems to be even if the top

autocrat has the right vision.

Related to the above, we contribute to a largely ignored area, that of reversals of

privatization in the literature of SOE reforms. This literature largely focuses on

privatization (Megginson and Netter 2001; Djankov and Murrell 2002; Estrin et al.,

2009). State sector still accounts for 39% of employment in Russia and 44% in China,

suggesting that privatization is far from complete and that there might be large-scale

renationalization. The recent political turmoil in Europe and the United States further

illustrates that other policies long considered important consensus among most

economists, e.g. free trade, may also face the possibility of being reversed.

2. Institutional Background and Hypotheses Development

Since late 1970s, the CCP has instituted a tournament-type promotion system,

which includes linking an official’s promotion to his/her performance in supporting

economic growth and maintaining social stability (‘harmony’) in a region (e.g., Bo,

2007; Li and Zhou, 2005). These policies are combined with mandatory retirement

ages for officials working at different levels, which have led to the changing of guards

from the top down every decade (e.g., Li, 1998). Mr. Jiang took over the highest post

in 1989 from Mr. Deng, and was the leader of the Party for the next decade. He then

handed over his position to Mr. Hu during the period of 2002-2004, who assumed his

post until the end of 2012. Given the secrecy of the transition of power within the

Party, ordinary citizens can only get a glimpse of the power struggles among the

different blocs, as witnessed by the rise and fall of Mr. Bo, Xilai.2

We follow research in political science and identify three main factions within

2 See, e.g., “China Drama Now a Murder Mystery”, Wall Street Journal, April 11, 2012; “In China, a Fall from Grace May Aid a Rise to Power”, New York Times, April 26, 2012; “People’s Republic of Scandal”, Time, May 14, 2012.

the Party by examining politicians’ resumes. The first is the Youth League Group, led

by former President and Party Secretary General Hu Jintao. His advancement to the

top began with the Secretary General of the Youth League (of the Communist Party)

in 1984. Officials in this group held top positions in the Youth League during earlier

stages of their careers, and worked in the Central Committee of the Youth League or

held the rank of provincial youth league vice secretary or higher. The second

bloc—the Shanghai Group—is led by Mr. Jiang Zemin, who became the party

secretary of Shanghai in 1985 before taking the post of the Secretary General of the

Party in 1989. Officials in this group all have work experience in the Shanghai

Municipal Government during 1985-1989. The third group is called the Princelings,

or the descendants of prominent and influential older generations of Party officials,

including both Mr. Xi, the current President and Secretary General, and Mr. Bo.

Both anecdotal evidence and political science research (e.g., Shih, Adolph and

Liu, 2012) show that personal connections and support by senior officials can help

advance an official’s career. The relationship among the three factions can be

characterized as ‘competition and cooperation,’ while officials not affiliated with any

of the factions are expected to have weaker support from senior officials, and thus

more concerned about protecting their status quo and not taking risks in setting

economic policies.

These political institutions affect the politician’s ability to handle sensitive issues

e.g. privatization. While China had completed the largest scale of privatization of

SOEs during much of the 1980s and 1990s, the privatization process is not always

smooth. Pursuing for-profit goals leads to large layoffs in many areas, as privatized

firms try to cut cost and increase labor productivity. When the local economy is

underperforming, unemployment may lead to social unrest (e.g., Zhu, 2005).

Government officials thus face tremendous pressure and may have to abandon or

reverse the privatization process. The following case exemplifies such problems.

In 2005, the provincial government of Jilin (in the northeastern part of China)

sold a 36% stake in Tonghua Iron & Steel, a SOE, to privately owned Jianlong

Corporation Group as part of restructuring effort. Jianlong also structured a deal with

the local government designed to increase its stake in Tonghua to a majority one.

Rumors began flying that Jianlong planned to build a new steel plant in another city

and replace current Tonghua workers with new recruits from there; the person of

interest—Mr. Chen, a 41-year-old executive of Tonghua and a representative from

Jianlong—was expected to execute plans of slashing jobs and shrinking pensions.

Thousands of workers who worried about losing their jobs staged a protest that shut

down production at the factory. A group of them found Mr. Chen and beat him

severely, fracturing his skull. Workers also blocked streets near the factory and hurled

bricks, preventing police and paramedics from reaching Mr. Chen. Later that night,

local government officials announced that the plan for Mr. Chen’s company to take

control of the steelmaker had been scrapped (Wall Street Journal, 07/31/2009).

These facts about China’s political system and privatization help us frame our

empirical tests. We are interested in how the structure of the political system affects

politicians’ policy choices. Politicians calculate their own benefits and costs and may

seek sub-optimal policies. In particular, the privatization of SOEs, despite its general

positive effect in improving efficiency, faces considerable constraints and incurs

social, political and economic costs, such as possible social instability caused by

unemployment. Thus, the privatization process is affected by political concerns of

local politicians. Without factional support, politicians might be more risk-averse and

are more sensitive to the pressure resulted from layoffs and possible social unrest.

Factional politics is a norm in many countries, including modern democracies

(Persico et al., 2007), and they are more common in the absence of full democracy

(Markevich and Zhuravskaya, 2011). We focus on party officials that hold the highest

position at the provincial level, i.e., provincial party secretary. China has 31 provinces,

offering a wide range of cross-sectional differences in economic development and

various dimensions of institutions. Under the supervision of the central government,

these provinces are regarded as fairly independent economic units and provincial

leaders are not involved with managing other provinces’ economies and policy setting.

Within a province, the party secretary, along with the governor, oversees the entire

economy and monitors officials at lower levels (e.g., cities and counties).3

The corporate event we look at is re-nationalization, defined to be local

governments (within a province) regaining ownership stakes in privatized firms. We

expect party officials without strong support and protection from senior leaders to be

more concerned about the short-run costs (e.g., unemployment) than the long-run

benefits (e.g., higher productivity) of privatization. Hence, we hypothesize that the

provincial party secretaries with no affiliation with any of the three factions are more

likely to support re-nationalization in their provinces so as to lower or stop rising

unemployment rates.

Our next set of hypotheses relate to the effects of re-nationalization on firm

performance. Based on the vast body of empirical evidence on privatization in China

and elsewhere, we hypothesize that re-nationalization has adverse effects on firm’s

performance as measured by profitability and labor productivity. Since

re-nationalization is not a random event, our main empirical strategy is to use an IV

model and a two-stage least square procedure (2SLS) to study the effects of

re-nationalization on performance. An official’s affiliation with a particular political

faction is either determined by his/her work experience with the supreme leader of the

party (Mr. Jiang and Mr. Hu) long before the official becomes the party secretary of a

province, or by his/her kinship to revolutionary heroes. Thus, officials’ affiliation

with the three factions, which measures the strength of their standings in the Party and

is linked to the likelihood of re-nationalization of firms in a region, should not affect

the performance of individual firms. Therefore, we use these indicators (among others)

as instruments for re-nationalization.

Theories of political economy stipulate that politicians can benefit from

economic policies even if they reduce efficiency. Politically driven economic policies

are not a new phenomenon in China: politicians, out of career concerns, can take

ill-conceived policies even if they may lead to dire consequences. Kung and Chen

3 In terms of overall standing among all party and government officials in China, provincial party secretaries and governors have the same administrative rank as those holding ministry-level positions (e.g., Minister of Finance, Minister of Justice) in the central government.

(2011) show that provincial politicians’ rank can explain a significant part of the high

death rates during China’s Great Famine, 1959-1961. They find excessively high tax

and procurement rates (and thus high death rates) in provinces governed by

lower-ranked officials, who had a stronger incentive to please the Party’s Central

Committee via contributions in the form of tax income and possessed goods. In our

context, one benefit of re-nationalization is lower unemployment rates, an important

indicator of “social harmony,” even if re-nationalization leads to worse performance

of the firms. Hence, in our final set of tests we explore the impact of unemployment

on the politicians’ promotion perspectives.

Overall, these tests will shed light on whether Chinese politicians could adhere to

a policy agenda, and the economic and political effects of their choices. The answers

to these questions will enhance our understanding of the working of an autocracy.

3. Data and Results

Our main data source is the Annual Survey of Industrial Firms (ASIF) released

by the National Bureau of Statistics of China for the period 1998-2007, which

contains all SOEs and non-SOEs with annual sales over RMB 5 million. Among all

the existing databases on Chinese corporate sectors, this is by far the most

comprehensive one for all types of firms with financial and accounting data. A firm’s

ownership type is updated each year in the database. In particular, for each state

ownership stake, the database specifies the owner’s name (i.e., which government

branch or agency) and administrative rank. We restrict our sample to firms that were

100% owned by local government as of the end of 1998 and went through a

privatization process afterwards. The firms that remain private after the initial

privatization process serve as the natural ‘control group’ for privatized firms that were

later ‘re-nationalized.’ We exclude firms that are owned by the central government:

provincial level officials have no control over these companies and our focus is on the

influence of local government officials on corporate sectors. We also drop firms

without sufficient information on key variables. Our final sample contains 4,727 firms

over the period 1998-2007.

A state-owned firm (as of 1998) is privatized if its state ownership stake drops

below 100% after 1998. After the initial privatization, there are three possible

outcomes for the firm in terms of ownership structure: 1) its state ownership stake

remains the same; 2) further privatization, i.e., state ownership continues to fall; and 3)

the state ownership stake increases (re-nationalization). Specifically, a firm is

re-nationalized if the state ownership stake increases following the initial

privatization.4 As such, 1,212 firms (25.6%) of our sample SOEs experienced a

reversal of privatization, and the average duration between privatization and

re-nationalization is about 2 years.

Figure 1 plots the frequencies of privatization and re-nationalization during the

sample period. We can see large-scale privatization took place in 1999 and early

2000s, and re-nationalization occurs as early as 2000 and continues throughout the

2000s. The magnitude of increases in government stakes is usually large: among the

re-nationalized firms, the average size of the state ownership increases from 43.2%

(post privatization) to 70.9% (after re-nationalization). In fact, local governments

regain the controlling position—ownership stake greater than 50%—in 441 firms; and

in 235 firms, local government stakes increase from 0 to 100%. Figure 2 plots the

ratio of the number of re-nationalized firms over incidences of newly privatized firms

over the sample period.

3.1 Univariate Comparisons of Renationalized and Control Group Firms

Table 1 compares re-nationalized firms and privatized SOEs that are not

re-nationalized. Data is obtained for the period after the initial privatization but before

local governments regain control of these firms. The average size of the state

ownership stakes of re-nationalized firms is 43.14% before the government takeover,

far greater than that of SOEs remaining privatized (17.33%). Re-nationalized firms

are larger, have higher leverage ratios, lower labor productivity and profitability;

4 We conduct robustness tests using more restrictive definitions of re-nationalization below (see Table 6). Our dataset does not provide information on the transaction prices through which local government acquired the ownership stakes.

these firms also employ more workers and account for a higher share of workforce in

local labor markets. Overall, these crude comparisons show that firms that are

re-nationalized by local governments appear to be less efficient and in worse financial

conditions than the rest of the privatized SOEs but play a more important role in the

local labor markets.

Table 1 also compares the privatization process of the two groups of firms in

Panel B. A common problem among all SOEs is low labor productivity in part due to

excessive work force. Hence, one of the major changes through privatization is the

reduction in labor force and enhanced productivity. We construct the variable ‘Layoff’

as Layofft = LnLt – LnLbefore_privatization, where Lt and Lbefore_privatization are the number of

employees in year t and one year before privatization, respectively. ‘Layoff’ thus

measures the percentage change in the number of workers since the initial

privatization. For re-nationalized firms, 10.8% of the work force is ‘released’ from

full-employment status after privatization; this figure is only half of the fraction of

workers laid off following privatization for the other firms (21.6%). Similarly, we

construct the variable ‘∆Sale’ to measure changes in sales. For re-nationalized firms,

sales increase by only 18.8% after privatization, as compared to an improvement of

30.2% for the other firms. We also find similar results for changes in labor

productivity with re-nationalized firms displaying a smaller improvement. These

comparisons further indicate that the privatization process for firms that are later

re-nationalized is not as complete as that of the other firms, and they do not

experience the same level of improvement in operation following privatization.

3.2 Provincial-level Political Structure and Its Relationship with

Renationalization

During the period of 1998-2007, there are a total of 75 provincial party

secretaries (across 30 provinces in China;5 a few officials hold the same post in

different provinces at different points of time), and 91 pairs of province-party

secretaries. The average tenure of a provincial party secretary is 5.58 years (median is

5 We drop Tibet due to the small number of firms located in the region.

5 years), with the shortest (longest) tenure 1 year (16 years). As discussed above, we

sort the officials into four groups: a) Youth League (18.4%), b) Shanghai Group

(17.7%), c) Princelings (5.8%), and d) not belonging to any faction (58.1%). Overall,

about 42% of the highest ranked provincial party leaders belong to one of the three

dominant factions.

As stated above, the affiliation with both the Shanghai Group and Youth League

Group is established when a top provincial leader worked in Shanghai during

1985-1989 or had working experience in the Youth League’s Central Committee in

Beijing or a provincial Youth League. We calculate the interval between the

establishment of such factional relationship of an official and the year when the

politician became the provincial party secretary. The mean and median of this interval

is 15.3 and 15 years for the Shanghai group, and 20.1 and 20 years for the Youth

League group.

We hypothesize that officials without any factional affiliation (and thus strong

political support) are more sensitive to issues that may jeopardize social stability, e.g.

unemployment. Therefore, they will have a stronger incentive to generate clear signs

of ‘social harmony,’ possibly at the expense of structural economic reforms. Figure 3

plots the number of provincial leaders among the four groups (three factions and the

‘no faction’ group). There is a downward trend in the proportion of provincial leaders’

without factional affiliation over our sample period. Not surprisingly, after Mr. Hu

took over the highest position of the Party in 2002, there had been a rise in the

number of provincial leaders belonging to the Youth League, while at the same time

the strength of the Shanghai Group declined following the retirement of Mr. Jiang

(the predecessor of Mr. Hu).

Panel C of Table 1 relates the frequency of re-nationalization (of firms in a

province) to the identity and association of the highest ranked official—the party

secretary in the province. In provinces where the party secretary has no ties to any

political faction, 9.29% of the privatized firms are re-nationalized. However, when the

official is affiliated with one of the three factions, only 5.58% of the privatized firms

are taken over by the local government (the difference between the two means is

significant at 1%). These results provide preliminary evidence on the influence of

factional politics and the status of the provincial party secretary on the likelihood of

local firms being re-nationalized.

3.3 Determinants of Renationalization

We first examine the determinants of firms’ re-nationalization by estimating the

following Probit model:

Pr (RNi,t =1) = Ф{φ0 +βFactionp,t + φ1Xi,t- 1+ φ2Zp,t + γt + ρp + ϛj + µip,t} (1)

where RNi,t takes the value of 1 if firm i is re-nationalized in year t, and 0 otherwise.

We drop all the observations for a firm after re-nationalization; we also drop

observations before the initial privatization, as the firm does not face the choice of

re-nationalization at that point. For example, suppose a firm is privatized in 1999 and

re-nationalized in 2003, then this firm has 4 observations entering the regression

model in Eq. (1), with RNi,t = 1 for 2003 and 0 for all four years during 2000-2002;

the observation in 1998 (first year of our sample period) is dropped. As a result of this

sampling procedure, we have a total of 15,266 firm-year observations, and the number

of unique firms is 4,727. We cluster standard errors by province so as to allow for

possible correlations among error terms from firms located in the same province.

Our key explanatory variable is the political status of the party secretary of the

province. We measure this by a dummy variable which indicates if he/she is affiliated

with one of the three main political factions within the CCP.

Vector X in Eq. (1) is a set of lagged firm controls, including firm size (log of book

assets), leverage ratio, profitability (return on sales, or ROS), the size of local

government’s ownership stake (State Share) and its squared term, labor productivity,

and the importance of the firm in the local labor market (Employment Share). We also

view the extent of labor force reduction following privatization (the variable “Layoff”

that measures the percentage drop in the labor force since privatization) as a factor for

re-nationalization. To control for the potential impact of political connection, we

include a variable that captures the administrative rank of the firm prior to its

privatization. In addition, we include a set of year, province, and industry indicators

(γt, ρp, ϛj) in all the models.

Table 2 reports the estimated marginal effects (evaluated at the mean of the

variables) of firms’ re-nationalization. First, we find that larger firms and firms in

worse financial conditions (higher leverage) and lower labor productivity are more

likely to be re-nationalized (statistically significant at the 5% or 1% level). For

example, according to the results of Column 1, as labor productivity drops by one

standard deviation, the likelihood of re-nationalization rises by 0.52 percentage points;

when leverage increases by one standard deviation, the probability of

re-nationalization increases by 1.61 percentage points. Firms’ profitability (ROS),

however, does not appear to affect their likelihood of re-nationalization.

Second, firms with larger local government ownership stakes (coefficient

significant at the 1% level in all the models), and firms that have not laid off a large

number of workers (significant at 5% in all the models) are significantly more likely

to be renationalized by local governments. For instance, when the layoff rate, or the

percentage of workers released from full-employment status after privatization (but

before re-nationalization), falls by one standard deviation, the probability of

re-nationalization rises by 1.02 percentage points. Given the unconditional probability

of a firm re-nationalized by the local government being 8.1% for the whole sample,

the result is also economically meaningful.

The above results suggest that the size of the labor force of the firm and the extent

of privatization play important roles in the government’s decision to re-possess the

control of privatized firms. Re-nationalization of these firms can have greater benefits

for the officials in that the re-nationalized firms can employ more workers thus help

stabilizing the local labor market. Further, the process of re-nationalization of these

firms can perhaps be accomplished at lower costs since the government still maintains

a large, albeit minority ownership stake of the privatized firms. Government stake has

a non-monotonic affect on the likelihood of re-nationalization: its squared term has a

negative sign (coefficient significant at the 1% level in all models), suggesting that

when the government holds an overwhelming position in the firm, there is no need to

further increase its ownership stakes.

The impact of factional affiliation. A key innovation of our study is to examine

the role of party officials in the organization of corporate sectors, and the results from

Table 2 confirm this role in driving renationalization. When a province’s party

secretary does not have close ties with any of the three dominant factions, firms

located in that province have a 1.64 percentage points higher chance to be

re-nationalized (Column 1, significant at 5%). Considering that the (unconditional)

probability of re-nationalization is 7.9% in the whole sample, this suggests that

political status and factional support of Party Secretary at the provincial level can

explain a large part of the re-nationalization process.

Omitted variables. In Column 2 of Table 2, we deal with the concern that the

estimates may be affected by the omitted circumstances faced by the politician. We

include a few variables that describe the institutional environment of the province.

First, a larger share of SOEs in the province indicates less private sector development

in the region. Second, the unemployment rate is a primary measure of social stability

in the province. Third, we include an index based on survey data from Fan and Wang

(2009) that measures the degree of government interference into corporate sectors; a

higher index indicates less government intrusion in the corporate sectors and better

protection of (private) property rights.6 We also include the growth rates of provincial

GDP.

The provincial level controls also come in as expected. For example, firms

located in regions with more government intrusion of corporate sectors (lower

Institution index), higher unemployment rates, and greater share of SOEs in the local

economies are more likely to be re-nationalized. If the Institution index drops by one

standard deviation—roughly equivalent to moving from Jiangsu, a developed coastal

province with a high institutional score, to Shaanxi, an underdeveloped inland

province with a low score, the probability of re-nationalization rises by 0.29

6 This is a component of the “NERI (National Economics Research Institute) Index” that measures economic development and institutional environment across provinces in China. Using the overall index yields similar results.

percentage points. These results are consistent with those firm-level factors,

suggesting that unemployment is a major concern for provincial officials, and these

officials are more likely to ‘interfere’ with private sectors if their general influence in

the local economy is greater. By contrast, GDP growth (prior to re-nationalization)

and fiscal status of the government is not related to the likelihood of

re-nationalization.

Finally, another factor that may affect re-nationalization is corruption: private

owners may collude with politicians when they sell their stakes to the government.

Although we do not have data on transaction prices, we try to control for this by

including a regional corruption index in column 2. As expected, a more corrupt

environment, measured by the frequency of exposed graft cases, is associated with a

higher likelihood of re-nationalization.

Overall, the results imply that the decision to re-nationalize is not made entirely

based on efficiency enhancement or profit maximization (e.g., for non-state

shareholders of the firms).

Hazard model. Another issue is whether the probit model is the right

specification. Alternatively we could use the duration model. The literature has shown

that the estimates from a discrete-time binary-choice model converge to those

obtained from a continuous-time duration model (Allison, 1982). Thus not

surprisingly, we reach qualitatively similar results in Columns 3 and 4 using the Cox

proportional hazard model

Difference between factions. To further investigate the impact of factional

politics, we run a regression similar to Eq. (1), with separate indicators for each of the

three factions. The default group is No Faction. Table 3 shows that being affiliated

with any of the three factions reduces the probability of renationalization, and that the

coefficient of the Princeling group is much larger than the coefficients of the other

two factions. This suggests that, possibly because of their inherited political capital,

politicians from high profile political families are the least likely to retreat from a

policy stance.

Ruling out Reverse Causality. There may be a concern of reverse causality of

our main results. We interpret the results as weak political status of the politician

leads to greater likelihood of firms being renationalized. Alternatively, one may think

that party officials without connections to the dominant factions are more likely to be

assigned to provinces with struggling privatized firms and worse labor markets. This

is quite unlikely based on the experience of President Hu Jintao: he was the party

secretary of Guizhou and Tibet, two of the most economically backward provinces,

before being promoted to Beijing. Nonetheless, to rule out this possible reverse

causality, we use an indicator variable that takes the value of 1 if the provincial party

secretary has no connection to any of the three groups, and 0 otherwise, as the

dependent variable in Table 4, and run a Probit regression with similar controls as in

Table 2, capturing firm-level performance (we use the average of each variable of the

firms located in a province) and provincial level variables. Based on the results

reported in Table 4, we do not see any of the firm-level or provincial-level variables

to be correlated with the political status of the party secretary. We conclude that

reverse causality discussed here is unlikely to explain our results.

More robustness checks. We conduct a number of robustness checks on the

determinants of re-nationalization. To save space, we only report the coefficients of

interest—the variables measuring the political status of the officials in Table 5. First,

since the frequencies of re-nationalization can be affected by different time trends

across regions, we add interaction terms between the year and regional indicators to

our baseline regression specified in Eq. (1). We continue to find similar results

(column 1). Second, we use different definitions of “renationalization” regarding the

threshold of increased government ownership stakes. More specifically, we consider a

firm to be re-nationalized in year t if the government share increases by at least 10%

(column 2) or if the government share increase from less than 50% to larger than 50%

(column 3). With these alternative definitions we obtain very similar results. Fourth,

we employ alternative definitions of factional affiliations. Following Shih et al.

(2012), we include an official’s social networks (e.g., alumni relations with the

faction leader) in addition to his/her working experience to determine factional

affiliation. We report qualitatively similar estimate is found, although the magnitude

is smaller (column 4).

Fifth, we also consider a narrower (and stricter) definition of political factions:

being affiliated with the current Secretary General of the CCP. Thus the “Shanghai

group” indicator equals one only if the politician is associated with Mr. Jiang Zemin

before 2002, when Mr. Jiang stepped down in 2002. The “Youth League group”

indicator equals one only if the politician is associated with Mr. Hu Jintao after 2002,

when Mr. Hu took presidency. This approach is essentially similar to Jia et al. (2015)

and Persson and Zhuravskaya (2016), where variation in the central leadership is used

to identify the strength of political connection of local politicians in China. Using this

new definition of factional affiliation, we continue to obtain similar results (column 5)

with weaker magnitude.

3.4 Economic and Political Effects

In Table 6 we examine the effects of re-nationalization on firm performance. We

first estimate the following fixed effect model:

Yi,t = β0 + β1Post-renationalizationi,t + γi + ρt + εit (2)

The dependent variables are Labor Redundancy (measured as the log of number of

employees over sales or over value-added), Profitability (return on sales), Total Factor

Productivity (including measures of TFP estimated with three methods), and

Leverage (total debt over total asset). Post-renationalizationi,t is a dummy variable

which equals one if a firm has been renationalized. γi and ρt are firm and year fixed

effects, respectively. The results are reported in Panel A.

In order to further control unobservable differences between firms which

experienced renationalization and other firms, we try to add more controls.

Specifically, we add industry-specific year fixed effects to allow for industry-specific

shock. To allow for different shocks between firms renationalized (treatment group)

and not renationalized (control group), we add group-specific year fixed effects. We

also add the interaction terms of initial firm-level characteristics7 and year dummies,

to control for the time effects of such characteristics on the outcome variables. The 7 Initial firm level characteristics are Firm Size, State Share, Leverage, Labor Productivity, Profitability, Layoff, and Employment Share in the year of privatization.

results are reported in Panel B.

As discussed before, re-nationalization is not an exogenous event, and hence we

also use a 2SLS procedure and an instrument to deal with the potential endogeneity

problem. It is unlikely that at the initiation of a factional relationship, an official

would choose the faction in anticipation that the leader of his/her faction would

eventually become the supreme leader of the Party many years later: the median of

the interval between an official’s initiation of a factional association and his/her first

year as a provincial party secretary is 15 years for the Shanghai group and 20 years

for the Youth League group. We thus use whether a provincial leader is affiliated with

a political faction as instrument.

We follow Wooldridge (2007) and fit a probit model with

Post-renationalizationi,t as the dependent variable, we then use the fitted probabilities

from this model as an instrument for Post-renationalizationi,t in a 2SLS estimation.

The probit model is as follows:

Pr (Post-renationalizationi,t =1) = Ф{ φ0 +φ1Xi,t-1 +φ2Zp,t + γt + ρp +ϛj + µip,t} (3a)

The first stage regression model (3a) is the same as our previous Probit model

specified in Eq. (1), except that the dependent vaible is Post-renationalizationi,t..

Similar to Eq. (1), independent variables include political faction status, firm-level

and province-level characteristics, industry dummies, year dummies, and province

dummies. We obtain the fitted value of Post-renationalizationi,t through estimating

(3a).

The we estimate the following linear regression model using the fitted value of

Post-renationalizationi,t as the actual instrument for the variable

Post-renationalizationi,t:

Yi,t= β0 + β1RNi,t + β2Xi,t-1 + β3Z’p,t + γt + ρp + ϛj + εip,t (3b)

The second stage (3b) is a linear regression model, with dependent variable Y the

change of performance metrics from one year before re-nationalization (t-1) to one

year after re-nationalization (t+1). The variables included in Zp,t in Eq. (3a) but

excluded from Z’p,t in Eq. (3b) are the indicators ‘no faction’. These provincial leaders’

characteristics serve as the determinants of a local firm’s re-nationalization. Finally,

we include the same set of firm-level and province-level controls in both stages, with

standard errors again clustered at the province level. The results are reported in

Panel C.

Testing for IV exclusiveness. As shown in Tables 2 and 4, factional affiliation,

which measures the strength of provincial leaders’ political status, affects the

likelihood of privatized firms being renationalized. Moreover, it should not (directly)

affect the performance of a specific local firm. In Table 4, we have already shown that

a provincial leader’s factional affiliation is not related to the (average) performance of

firms in the province. In addition, we also want to rule out other possible channels

through which officials’ political status can affect firm performance. For example,

connected politicians may receive more financial support from the central government,

which could in turn help improve firm performance in the province. To test this, we

regress the total amount of fiscal transfers a province receives from the central

government in a given year (scaled by the provincial fiscal revenue) on the provincial

leader’s political factional indicator and other controls in Table 7. Again, we do not

find that a politician’s factional affiliation affects the amount of transfers the province

receives from the central government. Based on the above tests, we conclude that the

political indicators satisfy the exclusion restriction, and serve as valid instruments for

re-nationalization in our study of the effects of re-nationalization on firm

performance.

Economic consequences. From Table 6, we observe that renationalization has a

negative impact on firm performance: the profitability and TFP of a renationalized

firm fall after the local government regains ownership stakes as compared to the year

before re-nationalization, than the control firms. Renationalization also raises labor

redundancy and firm leverage, which suggests that the government interferes with the

layoff process and by injecting more bank loans to the ill-performing firm. Putting

together the results, we have shown that renationalization is a sub-optimal economic

policy. The decision is made in part to protect the politician from being affected by

negative social consequences. While temporarily halting the trend of unemployment,

the economic impact of renationalization is negative and runs contrary to the central

government’s aim to maintain a high economic growth.

Political incentives. We have demonstrated that a politician’s factional affiliation

has profound impacts on his/her ability to adhere to an economic policy. To further

explore the political incentives behind this behavior, we examine the determinants of

promotion by estimating the following model:

Promotedi,t = β0 + β1Xi,t+ β2Zp,t + γp + ρt + εi,t (4)

where the dependent variable Promotedi is a dummy which equals one if the party

secretary i during the renationalization sample period 1999~2007 got a seat in the

Politburo by 2012, the recent National Congress of the CCP. The key explanatory

variables are the politician’s political status and economic indicators of the province.

We control for province fixed effects and year fixed effects. Column 1 of Table 8

shows that regional economic growth enhances the politician’s promotion prospectus,

consistent with the literature on career concerns of Chinese local officials. Not

surprisingly, unemployment works in the opposite direction. When we add the

politician’s political status to the regression in Column 2, we can see that politicians

with NO factional affiliations are disadvantaged in the promotion competition, after

controlling for economic growth and unemployment. Column 3 further disaggregates

factional ties into detailed classification and shows that being affiliated with one of

the three main factions helps one’s promotion, with the Youth League group

displaying a weaker effect. Growth and stability may have different impact on

different politicians, and we interact the factional affiliation with both GDP growth

and unemployment in Column 4. While the interaction term between GDP growth and

No Faction is insignificant, the interaction term between unemployment and No

Faction is negative and significant. The results suggest that high unemployment

would amplify the disadvantage of those politicians without strong factional support,

and that high economic growth may not mitigate the issue.

4. Conclusions

The Chinese government has played an important role in the transition from a

central planning economy to a mixture of market-based and government-controlled

sectors and has been hailed as a successful model of benevolent autocracy.Some have

even suggested that a “Beijing Consensus” is replacing the “Washington Consensus”

of the World Bank and IMF (Halper, 2010) partly because China boasts of “an army

of great ideas and successful implementation” (Ramo, 2004).

With a large sample of firms from thirty provinces across China during the period

1998-2007, we document, for the first time in the literature, large-scale reversals of

privatization: local governments regain ownership stakes in a quarter of previously

privatized firms. The reversal of the policy agenda is closely related to China’s

political structure: we find that party secretaries without strong support are more

concerned about protecting their status by introducing policies that can lead to

immediate, ‘desirable’ social outcomes such as lower unemployment rates at the cost

of long-run inefficiencies. Following prior research, we classify three dominant and

distinct factions within the Party that have been in existence for the past two decades,

and identify each provincial leader’s affiliation with the factions. We find that firms

located in provinces with party leaders who do not belong to any of the three

dominant political factions are more likely to be re-nationalized. Renationalization

leads to economic consequences that the Chinese government has been trying to avoid:

reduction in productivity and increases in labor redundancy and leverage. The

distortion fits well with the power structure in an autocracy: politicians without

factional affiliations experience a smaller likelihood of promotion and their political

fate hinges more on discouraging social unrest than on making economic achievement.

Our results shed new light on how political structure and politicians shape economic

policies, and in particular, how incentives of politicians can affect firms’ organization

structure and performance. By analyzing the determinants and consequences of

renationalization, we show that the seemingly efficient top-down system in an

autocracy is constrained by its Byzantine nature.

Going forward, many scholars have called for another reform in China, from the

state-led, investment intensive, export-oriented economy, which may have reached a

tipping point, to one relying much more on domestic consumption and private sectors’

production (see, e.g., World Bank, 2012). Under this new model, privately owned

firms will play an increasingly more significant role, and more supportive policies and

regulations are needed to help grow private sectors and allow them to compete with

SOEs on an even playing ground. However, the built-in distortions in China’s

autocratic system may explain the recent stagnation in its market-oriented reforms and

the massive build-up of debt.

References Acemoglu, D. and S. Johnson, 2000. “Unbundling Institutions”, Journal of Political

Economy 113, 949-995.

Acemoglu, D., S. Johnson, J. A. Robinson, and P. Yared, 2008, “Income and Democracy,” American Economic Review, 98 (3), 808–842.

Acemoglu, D., S. Naidu, P. Restrepo, and J. A. Robinson, 2017, “Democracy Does Cause Growth,” Journal of Political Economy.

Allen, F., J. Qian, and M. Qian, 2005. “Law, Finance and Economic Growth in China”, Journal of financial Economics 77, 57-116.

Allen, F., J. Qian, and M. Qian, 2008. “China’s Financial System: Past, Present and Future”, Ch. 14 in China’s Great Economic Transformation, L. Brandt and T. Rawski eds., Cambridge Univ. Press.

Allen, F., J. Qian, C. Zhang and M. Zhao, 2011. “China’s Financial System: Opportunities and Challenges”, in NBER book Capitalizing China, R. Morck and J. Fan eds., University of Chicago Press.

Allison, Paul 1982, “Discrete-time Methods for the Analysis of Event Histories,” Sociological Methodology, 13, 61–98.

Bentolila S. and G. Saint-Paul, 2003. “Explaining Movements in the Labor Share,"The B.E. Journal of Macroeconomics, 3(1), 1-33.

Bloom, N., R. Sadun, and J. Van Reenen, 2012. “The Organization of Firms across Countries.” Quarterly Journal of Economics 127(4): 1663-1705.Bo, Z., 2007. “China’s New Provincial Leaders: Major Reshuffling Before the 17th National Party Congress”, China: An International Journal 5(1), 1-25.

Bortolotti, B. and M. Faccio, 2008, “Government Control of Privatized Firms.” Review of Financial Studies 22, 2907–2939.

Bortolotti, B. and P. Pinotti, 2008, “Delayed privatization”, Public Choice, 136, 331-351

Boubakri, N., J. Cosset, O. Guedhami, and W, Saffa, 2011, “The Political Economy of Residual State Ownership in Privatized Firms: Evidence from emerging markets”, Journal of Corporate Finance 17, 244-258.

Cai, H. and D. Treisman, 2006. “Did Government Decentralization Cause China’s Economic Miracle.” World Politics, 58, 505-535.

Calomiris, C., R. Fisman and Y. Wang, 2010. “Profiting from government stakes in a command economy: Evidence from Chinese asset sales.” Journal of Financial Economics 96, 399-412.

Chernykh, L., 2011. “Profit or politics? Understanding Re-nationalizations in Russia”, Journal of Corporate Finance 17, 1237-1253.

Cohen, L., J. Coval, and C. J. Malloy. 2013. “Do Powerful Politicians Cause Corporate Downsizing?” Journal of Political Economy, 119, 1015–1060.

Cohen, L., and C. Malloy. 2014, “Friends in High Places.” American Economic Journal: Economic Policy 6(3), 63-91.

Dastidar, S., Fishman, R., Khanna, T., 2008. “Limits to Policy Reversal: Evidence from Indian Privatizations.” Journal of Financial Economics 89, 513–526.

Dinc, S., 2005, “Politicians and Banks: Political Influences on Government-Owned Banks in Emerging Markets”, Journal of Financial Economics 77, 453-479.

Dinc, S., and N. Gupta, 2011. “The Decision to Privatize: Finance and Politics”, Journal of Finance 66, 241-269.

Djankov, Simeon, and Peter Murrell. 2002. “Enterprise Restructuring in Transition: A Quantitative Survey.” Journal of Economic Literature 40(3): 739-792.

Estrin, S., J. Hanousek, E. Kocenda, and J. Svejnar, 2009. “Effects of Privatization and Ownership in Transition Economies”. Journal of Economic Literature 47(3), 699–728.

Fan, G., and X. Wang, 2009. NERI Index of Marketization of China’s Provinces 2009 Report (in Chinese). Beijing: Economic Science Press.

Fan, J., Wong, T. J., Zhang, T., 2007. “Politically Connected CEOs, Corporate Governance, and Post-IPO Performance of China’s Newly Partially Privatized Firms.” Journal of Financial Economics 84, 330–357.

Fisman, R., and Y. Wang, 2013. “Corruption in Chinese Privatization,” forthcoming, Journal of Law, Economics and Organization.

Gupta, N., 2005, “Partial Privatization and Firm Performance,” Journal of Finance 60, 987-1015.

Halper, S., 2010, The Beijing consensus: how China's authoritarian model will dominate the twenty-first century, Basic Books.

Huang, J., 2000, Factionalism in Chinese Communist Politics. Cambridge University Press.

Huang, Y., 2002, “Managing Chinese Bureaucrats: An Institutional Economics Perspective,” Political Studies 50, 61–79.

Huang, Y., 2008. Capitalism with Chinese Characteristics, Cambridge University Press.

Jia, R., M. Kudamatsu, D. Seim. 2015. "Political Selection in China: The Complementary Roles of Connections and Performance.” Journal of the European Economic Association, 13(4): 631-668

Kung, K. and S. Chen, 2011, “The Tragedy of The Nomenklatura: career incentives and political radicalism during China’s Great Leap Famine,” American Political Science Review 105(1), 27–45.

Lipset, Seymour M.,1959, “Some Social Requisites of Democracy,” American

Political Science Review, 53(1), 69-105. Markevich, A. and E. Zhuravskaya, 2011. “M-form Hierarchy With

Poorly-Diversified Divisions: A case of Khrushchev's reform in Soviet Russia,” Journal of Public Economics, 95(11-12), 1550-1560

Megginson, William, and Jeffry Netter, 2001. “From State to Market: A survey of empirical studies on privatization”, Journal of Economic Literature 39, 321-389.

Olley, S. and A. Pakes, 1996. “The dynamics of productivity in the telecommunications equipment industry,” Econometrica 64(6): 1263–1297.

Persico, N., J. Rodríguez-Pueblita, and D. Silverman, 2007. “Factions and Political Competition.” Journal of Political Economy, 119(2), 242-248.

Persson, P., E. Zhuravskaya. 2016. "The Limits of Career Concerns in Federalism: Evidence from China." Journal of the European Economic Association, 14(2): 338-374

Piotroski, J. and T. Zhang, 2013, “Politicians and the IPO decision: The impact of impending political promotions on IPO activity in China”, Journal of Financial Economics, forthcoming.

Rajan, Raghuram and Luigi Zingales, 1998, “Which Capitalism? Lessons from The East Asian Crisis”, Journal of Applied Corporate Finance, 11(3), 40–48.

Ramo, J.C., 2004, The Beijing Consensus. Foreign Policy Centre.

Sapienza, Paola, 2004, “The Effects of Government Ownership in Bank Lending”, Journal of Financial Economics 72, 357-384.

Shih, Victor, 2004. “Factions Matter: Personal Networks and the Distribution of Bank Loans in China”, Journal of Contemporary China, 13:38, 3-19.

Shih, Victor, 2008. ‘‘Nauseating’’ Displays of Loyalty: Monitoring the Factional Bargain through Ideological Campaigns in China, Journal of Politics, 70(4), 1-16.

Shih, V., C. Adolph and M. Liu, 2012. “Getting Ahead in the Communist Party: Explaining the Advancement of Central Committee Members in China”, American Political Science Review, 106, 166-187.

Shleifer, A., and R. W. Vishny, R. W., 1994. “Politicians and firms.” Quarterly Journal of Economics 109, 995–1025.

Song, Z., K. Storesletten, and F. Zilibotti, 2011. “Growing Like China,” American

Economic Review 101, 202-241.

U.S.-China Economic and Security Review Commission Staff Research Report, “The China Rising Leaders Project, Part 1: The Chinese Communist Party and Its Emerging Next-Generation Leaders”, 2012.

World Bank, 2012. China 2030: Building a Modern, Harmonious, and Creative High-Income Society, Joint with the Development Research Center of the State Council, the People’s Republic of China.

Zhu, X., 2005. “Misreading Law and the Imagined Homeland: a case study of the state-owned enterprise”, Chinese Sociology and Anthropology: A Journal of Translations, 38(1).

Table 1. Summary Statistics This table presents sample statistics of firm-level variables. Our main source of data is the Annual Survey of Industrial Firms (ASIF) released by the National Bureau of Statistics for the period 1998-2007. We restrict our sample to firms that were 100% owned by local government as of the end of 1998 and went through a privatization process, in which state ownership drops below 100%, afterwards. A firm is ‘re-nationalized’ if the state ownership stake increases following the initial privatization. The rest of the former SOEs remain privatized, and they serve as the ‘control group.’ Definitions of all the other variables are listed in Appendix A.

Privatized SOEs Privatized SOEs Mean Comparison Standard errors of Re-nationalized Not Re-nationalized of Two Groups of Mean Comps Panel A. Firm-level Variables State Share 0.4314 0.1733 0.2580*** 0.0078 Assets(million) 249.8616 196.6198 53.2418*** 19.228 Sales(million) 140.8838 147.5181 6.6342 19.0807 Labor Productivity 4.5020 4.8998 -0.3979 *** 0.0235 Profitability -0.0345 -0.0238 -0.0106* 0.0062 Leverage 0.7417 0.7228 0.0189*** 0.0072 Employment 895.8628 573.3529 322.51*** 35.5969 Employment Share 0.0522 0.0316 0.0205*** 0.0017 Political Connection 0.162 0.179 -0.0171** 0.0086 Panel B. Change of Firm Performance around Privatization ΔAssets 0.1154 0.0990 0.0164 0.0152 ΔSales 0.1878 0.3024 -0.1147*** 0.0188 ΔProductivity 0.3964 0.6229 -0.2265*** 0.0207 ΔProfitability -0.0049 0.0022 -0.0071 0.0084 ΔLeverage 0.0064 -0.0066 0.0130* .0071 Layoff 0.1079 0.2159 -0.1080*** 0.0118 Number of firms 1,212 3,515

Panel C. Policy Reversal and Political Factions This table compares the rates of firms’ re-nationalization across provinces. Our main source of data is the Annual Survey of Industrial Firms released by the National Bureau of Statistics for the period 1998-2007. We sort provincial leaders by whether they are affiliated with one of the three main political factions or not and whether they are new to office (the first year of their tenure). All the other variables are described in Appendix A.

No Faction Faction Comparing Means

Standard error of mean-comparison

Rate of re-nationalization 0.0929 0.0558 0.0371*** 0.0048 Number of firm-years 10,150 4,824

Note： *, ** and *** indicate statistical significance at 10%, 5% and 1% levels.

Table 2. Determinants of Policy Reversal This table reports Probit regressions on the determinants of firms’ re-nationalization. Firm-level data comes from the Annual Survey of Industrial Firms released by the National Bureau of Statistics for the period 1998-2007. The dependent variable takes on the value of 1 if a firm is re-nationalized in year t, and 0 otherwise. We report marginal probabilities evaluated at the mean of the variables. We include but do not report industry, province and year indicators. Standard errors are clustered at the province level, and t-statistics are in the brackets below the coefficients.

Probit Model Probit Model Hazard Model Hazard Model Firm-level Variables Log Assets 0.00375** 0.00380** 0.0539** 0.0543**

(2.293) (2.321) (2.285) (2.298) State Share 0.333*** 0.331*** 4.254*** 4.250***

(14.42) (14.44) (13.29) (13.36) Square of State Share -0.301*** -0.299*** -3.676*** -3.674***

(-12.74) (-12.77) (-11.71) (-11.77) Leverage 0.0176*** 0.0176*** 0.261*** 0.265***

(2.908) (2.894) (3.142) (3.172) Labor Productivity -0.00626*** -0.00636*** -0.0693** -0.0697**

(-2.759) (-2.811) (-2.096) (-2.102) Profitability -0.00379 -0.00381 -0.0466 -0.0491

(-0.577) (-0.583) (-0.615) (-0.658) Layoff -0.0135*** -0.0131*** -0.151** -0.146**

(-3.149) (-3.063) (-2.222) (-2.119) Employment Share 0.0334** 0.0339** 0.474** 0.494**

(2.059) (2.083) (2.216) (2.318) Political Connection -0.00688 -0.00666 -0.122 -0.121

(-1.075) (-1.044) (-1.291) (-1.278) Province-level Variables No Faction 0.0164** 0.0161** 0.247** 0.250**

(2.444) (2.461) (2.449) (2.511) Institution -0.00296* -0.0500**

(-1.907) (-2.319) Unemployment Rate 0.0118* 0.186**

(1.650) (2.054) Share of SOE 0.104** 1.268*

(2.032) (1.704) GDP Growth 0.0408 0.862

(0.180) (0.257) Fiscal Status -0.0796 -1.661

(-1.136) (-1.554) Corruption 0.496 9.597** (1.464) (2.067) Province, Industry and Year fixed effects

Yes Yes Yes Yes

Observations 14,886 14,886 14,886 14,886 Pseudo R-squared 0.124 0.125 0.0299 0.0305

Note： *, ** and *** indicate statistical significance at 10%, 5% and 1% levels.

Table 3. Differences between Factions We run a regression similar to column 2 of Table 2, replacing “No faction” in Table 2 with

three variables separately indicating the three factions. The default group is No Faction.

(1) Princeling -0.0316*** (-2.912) Youth League -0.0137* (-1.645) Shanghai Group -0.0194** (-2.056) Controls Yes Province, Industry and Year fixed effects

Yes

Observations 14,886 Pseudo R2 0.127

Note： *, ** and *** indicate statistical significance at 10%, 5% and 1% levels.

Table 4. Factional Affiliation and Firm Performance In this table we examine the possible relationship between provincial leaders’ political status and the performance of firms located in the province. The dependent variable, No faction, takes on the value of 1 if a provincial leader belongs to a political faction in year t, and 0 otherwise. The independent variables are the within-province averages of firm characteristics in year t. Standard errors are clustered by provinces, and t-statistics are in the brackets below the coefficients.

NO Faction Average Firm-level Variables State Share 0.0334

(0.399) Assets 0.827*

(1.646) Labor Productivity 0.251

(0.318) Leverage -0.0698

(-0.270) Profitability -0.881

(-1.444) Layoff -0.0171

(-0.0280) Employment share -0.620 (-0.849) Political Connection -0.0425 (-0.0887) Province-level Variables Institution -0.0182

(-0.803) Unemployment Rate -0.134

(-1.317) Share of SOE -0.345

(-0.575) GDP Growth 0.145

(0.0491) Fiscal Status 0.158

(0.243) Corruption 5.987

(1.070) Year/Province fixed effects Yes Observations 240 Pseudo R-squared 0.421

Note： *, ** and *** indicate statistical significance at 10%, 5% and 1% levels.

Table 5. Robustness Checks This table reports additional Probit regression results on the determinants of re-nationalization. The dependent variable takes on the value of 1 if a firm is re-nationalized in year t, and 0 otherwise. We report marginal probabilities evaluated at the mean of the variables. We include but do not report coefficients on firm-level and provincial-level controls, as well as industry, province and year indicators. Standard errors are clustered at the province level, and t-statistics are in the brackets below the coefficients. In Column 1, we add interaction terms between the year and regional (western, middle and eastern) dummies to our baseline regression specified in Eq. (1), in order to control possible region-specific trends. In Column 2, a firm is re-nationalized in year t if the government share increased by at least 10%. In Column 3, a firm is re-nationalized in year t if the government share increased by less than 50% to more than 50%. In Column 4, No Faction takes on the value of 1 if a provincial leader belongs to a political faction as classified by Shih et al. (2012). In Column 5, No Faction takes on the value of 1 if a provincial leader belongs to a strong political faction.

(1) (2) (3) (4) (5) Control for

region-year FE

Alternative definition I of renationalization

Alternative definition II of renationalization

Alternative definitions of factional ties I

Alternative definitions of factional ties II

No Faction 0.0217*** 0.00385** 0.0141** 0.0119* 0.0126* (3.316) (2.114) (2.232) (1.875) (1.939)

Controls Yes Yes Yes Yes Yes Year, Province and Industry FE

Yes Yes Yes Yes Yes

Region×year FE Yes Observations 14,886 15,297 15,714 14,886 14,886 Pseudo R2 0.126 0.152 0.105 0.123 0.123

Note： *, ** and *** indicate statistical significance at 10%, 5% and 1% levels.

Table 6. Effects of Renationalization on Firm Performance We report the impact of re-nationalization on firm performance. The variable Re-nationalized, which is 1 if the firm is re-nationalized, is instrumented by the fitted probabilities obtained from a first-stage Probit model, which has Re-nationalized as the dependent variable and is reported in Panel B. The dependent variables in Panel A are changes in profitability, productivity and employment from one year before to one year after re-nationalization. We use three methods to estimate total factor productivity (TFP). The first is estimated by ordinary least squares (OLS). We call this TFP-OLS. The second is by the method proposed by Olley and Pakes (1996). We call this TFP-OP. The third method relies on the estimates of the factor shares at the two-digit industry level from Bentolila and Saint-Paul (2003), as in Bloom et al. (2012). We call this TFP-Index number. The control group is privatized firms that have not been re-nationalized during that period. In Panel A, we also report first stage F-test statistics of excluded instruments of the first stage probit regression to validate that the fitted probability is a strong IV for Re-nationalized. The variables in the first-stage Probit model but excluded from the performance regressions (in Panel A) are the political variables. Firm-level controls are: Firm Size, State Share, Leverage, Labor Productivity, Profitability, Layoff, and Employment Share. Provincial-level controls：Government interference，Unemployment Rate，Share of SOE，GDP Growth and Fiscal Status. Standard errors are clustered at the province level.

(1) (2) (3) (4) (5) (6) (7) VARIABLES Labor/sale Labor/value-adde

d Profitability TFP OP TFP OLS TFP Index

number Leverage

Panel A: control for firm FE and year FE Post-renationalization

0.0535** 0.0309 -0.0160* -0.0477** -0.0408* -0.0439* 0.0210*** (2.498) (1.263) (-1.844) (-1.970) (-1.665) (-1.810) (3.412)

Firm FE YES YES YES YES YES YES YES Year FE YES YES YES YES YES YES YES Observations 21,866 21,083 21,902 21,034 21,050 21,050 22,030 R-squared 0.780 0.724 0.535 0.716 0.712 0.705 0.750 Panel B: control for firm FE, Industry FE×Year FE, Group Dummy×Year FE, and Initial firm-level characteristics×Year FE Post-renationalization

0.0745*** 0.0674** -0.0198* -0.0827*** -0.0853*** -0.0806** 0.00184 (2.752) (2.128) (-1.772) (-2.614) (-2.664) (-2.551) (0.229)

Firm FE YES YES YES YES YES YES YES Industry×Year FE YES YES YES YES YES YES YES Group Dummy×Year FE

YES YES YES YES YES YES YES

Initial firm-level characteristics×Year FE

YES YES YES YES YES YES YES

Observations 21,783 21,086 21,823 21,005 21,019 21,019 21,899 R-squared 0.792 0.735 0.567 0.723 0.719 0.713 0.753 Panel C: IV estimates Post-renationalization

0.110* 0.249** -0.0870** -0.182* -0.222** -0.183* 0.0674* (1.663) (2.515) (-2.560) (-1.739) (-1.977) (-1.754) (1.830)

Controls YES YES YES YES YES YES YES Observations 21,081 20,421 21,082 20,245 20,245 20,245 21,065 R-squared 0.757 0.523 0.295 0.440 0.362 0.422 0.104 First-Stage Probit on Re-nationalization No Faction 0.0187*** (2.733) Controls Yes Observations 21,902 R-squared 0.199

Note： *, ** and *** indicate statistical significance at 10%, 5% and 1% levels.

Table 7. Exclusiveness Check: Factions and Resources

(1) (2) VARIABLES Dep. Var.: fiscal subsidy/own fiscal revenue NO faction 0.00412 0.00366 (0.658) (0.614) GDP per capita -0.192*** (-3.650) Unemployment rate -0.00197 (-0.260) Share of SOE -0.114*** (-3.421) Province and Year fixed effects

Yes Yes

Observations 248 248 R-squared 0.975 0.978

Table 8. Factions and Promotion This tables reports factors that impact politicians’ promotion. The dependent variable is a dummy which is one if the politician gets a seat in the politburo or the standing committee of the politburo.

(1) (2) (3) (4) Unemployment Rate -0.0928* -0.0917* -0.100** 0.0163 (-1.833) (-1.891) (-2.044) (0.239) GDP Growth Rate 2.882** 2.845** 2.889** 1.166 (2.227) (2.318) (2.352) (0.635) Princeling 0.167** (2.128) Youth League 0.0861* (1.793) Shanghai Group 0.221** (2.503) No Faction -0.123*** 0.144 (-3.101) (0.521) No Faction*Unemployment -0.144**

(-2.550) No Faction*GDP Growth 2.407

(1.446) Province and Year fixed effects Yes Yes Yes Yes Observations 248 248 248 248 R-squared 0.763 0.794 0.796 0.802 Note： *, ** and *** indicate statistical significance at 10%, 5% and 1% levels.

Figure 1. Incidences of Re-nationalization

The graph displays the number of firms that were privatized or re-nationalized in each year during 1999-2007. Firm-level data comes from the Annual Survey of Industrial Firms released by the National Bureau of Statistics. Figure 2. Incidences of Re-nationalization over Incidences of Privatization

The graph displays the ratio: the number of firms that are re-nationalized / the number of firms that are privatized in each year over the period 2000-2007.

050

01,

000

1,50

02,

000

1999 2000 2001 2002 2003 2004 2005 2006 2007

Incidences of Renationalization Incidences of Privatization

0.29

0.36

0.46

0.63

0.46

0.63

0.57

0.43

0.00

0.20

0.40

0.60

0.80

2000 2001 2002 2003 2004 2005 2006 2007year

Figure 3. Political Factions of Provincial Leaders

The graph displays the number of provincial leaders that belong to a certain political faction, using our own definition, each year.

0.2

.4.6

.8

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Youth League Group Shanghai GroupPrincelings No Faction

Appendix A. Variable Definitions and Sources

Variable Description Sources

Re-nationalized Dummy variable which takes on the value of 1 if a firm is re-nationalized in year t, and 0 otherwise. ASIF

No faction Dummy variable which equals one if the provincial party secretary does not belong to a faction of CCP. Hand-collected

Assets Annual assets in thousands of Yuan. ASIF Sales Annual sales in thousands of Yuan. ASIF Profitability Annual profit over annual sales. ASIF Labor Productivity Annual sales per employee (in Log) ASIF

Employment Log of the number of a firm’s employees. ASIF Leverage The debt-to-asset ratio. ASIF State Share Percentage of government stake in the firm. ASIF

Employment Share

The firm’s number of employees over the local county’s total manufacturing labor force. ASIF

Layoff Reduction in the firm’s employment after privatization (in Log). ASIF Political Connection

Dummy variable that takes one if the firm was previously under the direct control of provincial government. ASIF

Institution Constructed from survey data on the time spent by the firm management dealing with government officials; a higher index indicates less government interference in the corporate sector.

NERI-China Marketization Index

Unemployment Rate Annual provincial unemployment rate. China Statistics

Yearbooks

Share of SOE The number SOEs as a fraction of total number of firms in a province. ASIF

GDP Growth The growth rate of provincial GDP. China Statistics Yearbooks

Fiscal Status Transfers received from central government over expenditures of the provincial government.

China Statistics Yearbooks

Corruption Graft cases filed per thousand people. Procuratorial Yearbooks of China

Appendix B: Construction of the panel from the ASIF data

Construction of the panel from the ASIF data. In the dataset, every firm is given a unique firm code. A small number of firms may have changed their firm codes within the sample period but remained in the sample. To address this issue, we follow Brandt et al. (2012) and Yang (2015) to obtain unique firm codes based on the firm’s name, zip code, telephone number, and founding year. We clean the data as follows. First, if the year t observation of a firm cannot be matched to any firm’s observation in year t+1 based on the firm code, we try to find a firm with the same name in year t+1, and match them by giving the year t+1 observation the same firm code as the year t observation. Second, for those firms that cannot be matched by the code or name, we rely on the combinations of the zip code, telephone number and the founding year to match them. We delete firms with missing key information, i.e. assets, fixed assets, sales and employment. Table A-1 presents the frequency with which we can link the observations in different years for both SOEs and non-SOEs.

Table B-1. Evolution of the raw panel over time Year Total firms Entrants Incumbent, linked using Exiting (in the next

year) NBS ID Other information

1998 164,452

28,709

1999 161,439 25,696 130,863 4,880 27,672

2000 162,350 28,583 130,538 3,229 36,395

2001 170,780 44,825 117,526 8,429 24,356

2002 181,149 34,725 142,950 3,474 28,378

2003 196,204 43,433 146,605 6,166 51,295

2004 274,750 129,841 137,681 7,228 45,085

2005 271,819 42,154 226,675 2,990 25,819

2006 301,943 55,943 243,728 2,272 28,485

2007 336,742 63,284 271,629 1,829

Note: Entrants are those that first appear in the sample in the specific year. Exiting means dropping out of the sample in the next year. The ASIF dataset includes all SOEs, and all non-state firms with sales exceeding five million yuan. Thus, a firm's entry year may be different from its establishment year. Similarly, a firm’s exiting year may differ from its death year.

References:

Brandt, Loren, Johannes Van Biesebroeck, and Yifan Zhang, 2012. “Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing.” Journal of Development Economics 97(2): 339-351.

Yang, Rudai, 2015. “Study on the Total Factor Productivity of Chinese Manufacturing Enterprises” (中国

制造业企业全要素生产率研究). 经济研究 (Economic Research Journal, in Chinese) 2: 61-74.